Genetic Analysis of Social Responsiveness and Head Circumference as Endophenotypes for Autism Spectrum Disorders

Background:  Extensive genetic heterogeneity in the autism spectrum disorders (ASD) has been challenging in linkage and association studies to date.  Increased head circumference (HC) is a prominent physical feature in ASD patients, with estimates of macrocephaly ranging from 12-30%.  The Social Responsiveness Scale (SRS) is a quantitative assessment of social ability, one of the three core deficits of ASD.  We sought to reduce phenotypic heterogeneity in ASD by using HC and SRS as quantitative endophenotypes for genetic analyses.

Objectives:  To identify genetic susceptibility loci for ASD using linkage and association approaches for the analysis of social responsiveness, and of ASD in patients and family members with above-average head circumference.

Methods:  We examined head circumference (HC) in the Autism Genetics Research Exchange (AGRE) cohort and adjusted for age and gender according to standard growth charts.  Nuclear families in which two or more offspring exhibited above-average HC were selected for linkage analysis.  

Raw SRS scores for patients with autism and their unaffected siblings were obtained from the Autism Genetics Research Exchange (AGRE).  We assessed correlation between SRS scores submitted by the subjects’ teachers and those submitted by parents and elected to use scores primarily from teachers, supplementing with scores from parents when necessary.  Age was not a significant covariate.  Although gender was a significant covariate, we elected to use raw SRS, as gender differences in SRS are likely confounded by the male:female ratio in patients with autism.  

Genome-wide SNP data for AGRE subjects were analyzed for linkage and association to ASD or to SRS as a quantitative trait.  Following SNP quality control, a subset of independent markers was selected (pairwise r²≤0.1) for non-parametric linkage analyses in Merlin.  Regions of interest were identified based on peak LOD scores, and additional SNPs in those regions were analyzed for association to ASD.

Results:  We identified significant linkage for ASD on chromosome 8q21 (LOD 3.21; rs6994256) in families with above-average HC.  The one-LOD interval surrounding the linkage peak spans ~6.7 Mb and includes 24 known genes.  We identified additional markers under the linkage peak on chr8q21 and report results for tests for association in the presence of linkage.

Quantitative linkage analysis for SRS showed a suggestive peak on chromosome 4q13.1 (LOD=2.52; rs1124974).  The one-LOD interval surrounding the linkage peak spans 11 Mb and 19 known genes, including EPHA5, which is thought to function in synaptic remodeling.  Using additional markers in the linked region, we observe an excess of association signal in the linked region for ASD, but not SRS.  We also note that Weiss et al (2009) reported suggestive association with ASD for SNPs near EPHA5 (p=8.5x10^-6; rs17088254).  Because our cohort overlaps with that analyzed by Weiss et al, the correlation in association findings cannot be considered independent evidence.  However, our linkage and association results are consistent in suggesting the existence of ASD susceptibility loci on chromosome 4q13.1.

Conclusions:  Linkage and association studies of HC and SRS support the utility of quantitative traits related to ASD in improving our ability to identify genetic susceptibility loci.